Saturday, 10 August 2013

Jason Richwine and some Hispanic data


It was in a Woody Allen film that our earnest hero was waiting in line to watch a movie while having to listen to a poseur in front of him talking nonsense about the concept of “the medium is the message”. Frustrated at hearing this idea being mis-represented, Woody turns behind him, and gets the progenitor of the idea, Marshall McLuhan, to step forward and give an authoritative explanation of what he really meant, thus putting the braggart in his place. Woody turns to the camera and says “If only real life were like this”.

It just so happens I can help correct some of the views being expressed by random commentators on the Jason Richwine affair. Heiner Rindermann and I have written a paper on this very topic of scholastic achievement which is online at Intelligence. The paper version will be out in December, but here is a preview.

Ability rise in NAEP and narrowing ethnic gaps? Heiner Rindermanna, Corresponding author contact informationJames Thompsonb E-mail the corresponding author


US National Assessment of Educational Progress (NAEP) results from 1971 to 2008 enable four different effects to be distinguished: Cohort rise effects, gap-narrowing between ethnic groups, trends due to demographic changes in by NAEP listed or not listed ethnic groups. NAEP means and percentiles in reading and mathematics were transformed to conventional IQs and SDs. The total increase from 1971 to 2008 was in the scale of 4.34 IQ points (dec = 1.17 IQ per decade). The ability distribution became more homogenous (down from SD = 15.00 in 1971 to SD = 13.56 in 2008). Increases were larger for youngerstudents (9-year olds: 2.02 IQ per decade; 13-year olds: 1.20; 17-year olds: 0.30); larger at the lowerability level (10th percentile dec = 1.79 vs. 90th percentile dec = 1.03). The largest increase was for Blacks (Whites dec = 1.29 IQ, Hispanics 2.27, Blacks 3.04). White–Hispanic-differences were reduced from 11.59 to 8.46 IQ, White–Black from 16.33 to 9.94 IQ. If the racial composition of the population had not changed, the mean gain for the 17-year-old group would have been 2.47 IQ points higher. Had the gap between Whites and the two other groups not narrowed, the mean gain would have been 1.70 IQ points lower. Demographic change has accounted for a loss of 2.47 IQ points and according to cognitive human capital theory $2001 GDP per capita per year, but total ethnic gap-narrowing has provided a gain of $1377.


  • Intelligence;
  • FLynn effect;
  • White–Black-differences;
  • Human capital


1-s2.0-S0160289613000895-gr1 ethnic gaps


The snapshot shows that the gap between the blue line for European Americans and the red line for Hispanic Americans narrowed in the early 1980s, but remains substantial. Same picture for African Americans.

In summary, if one uses scholastic data as a measure of ability, then there was some significant narrowing of the White/Hispanic gap in the early 1980’s, but that reduced gap has persisted thereafter. This finding discomforts those who predicted that the gap would never change, and those who said it was closing fast and would shortly disappear. Both are wrong, are partially right. Funny thing, facts. The same happened to African-American scholastic achievement, which had been the focus of much attention, Hispanic achievement somewhat less so.


  1. There's some sort of editing error here: " ... enable four different effects to be distinguished: Cohort rise effects, gap-narrowing between ethnic groups, trends due to demographic changes in by NAEP listed or not listed ethnic groups."

    1. Four different effects:
      1. Cohort rise effects;
      2. gap-narrowing between ethnic groups;
      3. trends due to demographic changes in by NAEP listed ethnic groups;
      4. trends due to demographic changes in by NAEP not listed ethnic groups.

  2. Interesting that blacks reached parity with Hispanics. Any idea why? That seems rather anomalous: usually Hispanics score significantly lower.

  3. It would be better to write "... four different effects to be distinguished: cohort rise effects, gap-narrowing between ethnic groups, and trends due to demographic changes in both NAEP-listed and NAEP-unlisted ethnic groups."

    It would be even better if you could rephrase that first effect to avoid the duplication effect of effects/effects, but that may be inconvenient.

    P.S. What are "cohort rise effects"? Is that an allusion to the Flynn effect?

    1. Yes, an allusion to the Flynn effect!
      I am no native speaker but I hope to continously improve my English!
      Heiner Rindermann

  4. @Prof Rindermann, I do hope that my remark didn't upset you; it was intended to be helpful. Anyway, your English is infinitely better than my German, which has decayed to almost zero.

  5. I was the one who used the term "cohort effects" because this was proposed by Earl Hunt, who has always been clear that the effect was well known in ageing research, long before psychometricians found it. So, it was a deliberate attempt to avoid the argument about who found it first: cohorts vary, and that is worth studying. By the way, when I post Jim Flynn's comments on the special issue, it will become clear he did not seek to be named as the discoverer of the cohort effect.

  6. At the risk of flogging a dead horse, I should congratulate you both for using "effects" rather than the modish, tedious and inaccurate "impacts". Or perhaps I should congratulate your whole discipline?